As a lecturer at the University of Wisconsin-Milwaukee’s School of Information Studies, Matthew Friedel often hears the question, “Will robots take my job?”
Not necessarily. But artificial intelligence (AI) will eventually disrupt around half of all human occupations, Friedel said Thursday in a tech panel held at the Overture Center.
Iya Khalil, head of the global AI innovation center Novartis, and Anthony Gitter, a University of Wisconsin-Madison professor of biostatistics and medical informatics, joined Friedel for a discussion on the use of AI in medicine at this year’s Wisconsin Biohealth Summit. The three speakers shared how companies, including Wisconsin’s postsecondary institutions, are using technology to improve human health.
“This time right now is almost analogous to the birth of the web, or the birth of mobile and social media,” Friedel said, referencing the boom in businesses creating AI. “One of the roles I have at the university is to identify these spaces so that we can educate our students for the jobs going forward.”
Friedel explained that AI — machine and computer simulations of human intelligence processes — has two characteristics: autonomy and adaptivity. AI can perform tasks without constant guidance from a user, allowing machines to think like people. Machine learning, an application of AI, trains systems to automatically learn and improve from experience without being explicitly programmed.
“At the root of machine learning and artificial intelligence is trying to extract knowledge from our data,” Friedel said.
Madison is already at the forefront of AI as a growing tech hub, according to the Brookings Institution. In its recent analysis examining “the extent, location and concentration” of AI in U.S. metropolitan areas, Brookings listed Madison as one of 21 cities that spent significant federal dollars on AI research and development.
At UW-Madison, Gitter leads a lab through the Morgridge Institute, which uses AI in drug discovery and disease treatment. He said the university has made significant strides in developing this tech to create custom-fit chemicals — or as he described it, “a brand new recipe” for treating illnesses.
When the lab of students, faculty and researchers began studying how to make new bacteria-killing chemicals, and after testing 400,000 of them, Gitter said the work only resulted in .1% effectiveness. The other 99.9% of the substances “completely failed,” he said.
With the help of chemists and another campus tech lab, the researchers trained machine learning models to “explore a billion more recipes that are commercially available.”
“We came up with a small list of the 68 chemicals that look very appealing for machine learning criteria,” Gitter said. “Because the system is guiding us about what to test, we can have a much more customized view of which chemicals actually work.”
Machine learning improved at identifying these new chemicals, increasing from .1% to nearly 50% effectiveness. New findings on the research project will be released in the coming weeks, Gitter said.
Khalil also spoke of her work in Wisconsin, performing drug trials across multiple sclerosis, psoriasis and a number of cancers. AI is getting stronger and “really kinetic,” she said, with the ability to drastically improve people’s lives.
She pointed to tech as a way to bring better treatments and diagnoses to patients with illnesses, calling on scientists to reimagine the way they use data.
While Friedel added that humans can leverage “creativity, improvisation, dexterity, judgment and social leadership” in ways AI cannot, machines are more useful for accuracy. Combining the two is where “the power of AI comes in.”
“That will cause disruptions,” Friedel said. “But it’ll also give us the ability to have humans work for higher economic value. That’s really the key piece.”
